Stanford Pattern-based Information Extraction and Diagnostics (SPIED) is a pattern-based entity extraction and visualization that provides code for two components, Learning entities from unlabeled text starting with seed sets using patterns in an iterative fashion and Visualizing and diagnosing the output from one to two systems.
Customer Reviews
Lissette L.
Advanced user of Stanford SPIEDSPIED can help us compare 2 systems by their identities. It's very intuitive, as it learns as you go.
SPIED visualizes learned patterns and entities, and also has the ability to contrast them.
Pattern learning is very important in my workflow, and this is a great way to do it.
There are some improvements to be made, it lacks a feature for users to provide
their own oracle label of a learned entity using SPIED. Also, visualization by parameters would make it easier for the user to understand systems that extract large number of patterns and entities.
Keep in mind that SPIED is still in the works, and they are improving features.
In the future I hope that they implement visualization by summarizing the diagnostic information.
However, if you are looking for a light web browsing experience, this is the way to go.
Also, SPIED-Learn uses TokensRegex to create and apply surface word patterns to text. It also inputs in a JSON format. Javascript, angular, and jquery are required to visualize information in a web browser.
SPIED helps me diagnose errors and tune parameters in a pattern-based system. It's also quite easy and efficient way. Also, let's not forget that it's a light-weight web browser based visualization and that's a plus in my workflow.
